From Smithian Growth to Schumpeterian Development: An Inquiry into the Development of the Kiryu Weaving District in the Early 20th Century Japan Tomoko Hashino* and Keijiro Otsuka** 25 November, 2011 Abstract This study finds that the process of evolutionary development of the Kiryu weaving district in Japan from 1895 to 1930 can be divided into the two phases, i.e., Smithian growth based on the inter-firm division of labor using hand looms and Schumpeterian development based on factory system using power looms. Weaving manufacturers-cum-contractors led Smithian growth by organizing sub-contracts with out-weavers in rural villages among others, thereby contributing to the steady growth in production. Newly emerged joint stock firms played a role of genuine entrepreneurs by realizing significant scale economies and transforming the traditional weaving district into a cluster of large modern factories. Keywords: industrial district, Smithian growth, Schumpeterian development, weaving industry, 20th century Japan __________________ * Graduate School of Economics, Kobe University. Email: [email protected]** National Graduate Institute for Policy Studies. Email: [email protected]
Transcript
From Smithian Growth to
Schumpeterian Development: An Inquiry
into the Development of the Kiryu Weaving
District in the Early 20th Century Japan
Tomoko Hashino* and Keijiro Otsuka**
25 November, 2011
Abstract
This study finds that the process of evolutionary development of the Kiryu weaving
district in Japan from 1895 to 1930 can be divided into the two phases, i.e., Smithian
growth based on the inter-firm division of labor using hand looms and Schumpeterian
development based on factory system using power looms. Weaving
manufacturers-cum-contractors led Smithian growth by organizing sub-contracts with
out-weavers in rural villages among others, thereby contributing to the steady growth in
production. Newly emerged joint stock firms played a role of genuine entrepreneurs
by realizing significant scale economies and transforming the traditional weaving
district into a cluster of large modern factories.
* Graduate School of Economics, Kobe University. Email: [email protected]
** National Graduate Institute for Policy Studies. Email: [email protected]
1
1. Introduction
Pioneering studies on industrial districts or clusters in business and economic
history by Piore and Sabel (1984) and Sabel and Zeitlin (1997) have contributed to clear
understanding of their important roles in the development of national economy in
Western countries. The investigation of industrial districts per se was not new; their
studies were new because they tried to explain major advantages of industrial districts
by using the concept of ‘externalities’ which Alfred Marshal introduced almost one
century ago (Marshall 1920). In fact, by analyzing industrial districts through the lens
of such externalities, the nature of competition and source of competitive advantage
have been more clearly identified (Porter 1998).
By reviewing the burgeoning literature on industrial districts or clusters in
many countries, regions, and industries, Zeitlin (2008) concludes his article by
highlighting three major remaining research questions; (1) the relationship between the
district and the wider world, (2) the changing morphology of the districts and
relationships among different sizes and types of firms within them, and (3) governance
and coordination mechanisms within the districts. In the case of Japan, recent studies
on industrial districts focus mainly on the last point (Abe 1992, 1999; Fujita 1998;
Hashino and Kurosawa 2011; Tolliday and Yonemitsu 2007). Above all, collective
2
institutions and organizations within the district played an important role in the
introduction and diffusion of new technologies, as they entailed technology spillovers
among firms and created the problem of inferior quality products, which damaged the
reputation of the district (Sawai 1999; Hashino forthcoming).1
In contemporary developing world, cluster-based industrial development is
widely observed. Sonobe and Otsuka (2006, 2011) primarily analyze the determinants
of the quality improvement of products and the possibility of exports from the clusters
in Asia and sub-Saharan Africa, which correspond to the first problem identified by
Zeitlin. Nadvi (1999) and Shumitz (1995) discuss the governance and coordination
mechanism of the cluster, which lead to what they call collective efficiency in the
context of South Asia and Latin America. Their studies squarely address the third
issue of Zeitlin. While the importance of industrial clusters for economic development
has been well analyzed by them, the historical or long-term perspectives are limited in
their studies.
The aim of this study is to explore how and why the different sizes and types of
firms within the district appeared, grew and collapsed in the long-term development
1 Hashino (2007b, 2010) explores Zeitlin’s first and second points. The former study analyzes the relationship between small-scale firms within the district and large-scale ones located outside. The latter study attempts to clarify how newly-developing weaving districts solved the problem of inferior quality products, which reduced the districts’ reputation at the international markets.
3
process of Kiryu weaving district in the early 20th century. Kiryu was one of the most
advanced silk weaving districts in the Tokugawa period (1603-1868). It was a pioneer
in export of silk products in the 1870s as well as the leading producer of traditional
Japanese kimono and obi for domestic markets. In this study, we will demonstrate that
three different types of players attempted to lead the growth of Kiryu. The first is
relatively large firms established in the late 19th century which introduced the vertically
integrated production system for mass production of export products. The second is
who promoted division and specialization of labor with village-based out-weavers and
other specialized small firms. Putting-out system for weaving, dyeing, and preparatory
and finishing processes prospered in the early 20th century. The third is joint stock
firms established in the early 1910s which adopted power looms and successfully
sought the scale economies, thereby “destroying” the out-weaving system and
“constructing” factory system. Following Parker (1984) and Mokyr (1990) who study
the historical patterns of economic change in the Western world, we would like to
demonstrate that Kiryu experienced Smithian growth based on the expanded division of
labor among a large number of firms, followed by Schumpeterian development leading
to the destruction of out-weaver systems and the creation of factory systems. We also
4
inquire into the causes of the success and failure of the three types of weaving firms.
The rest of the paper is organized as follows. The next section describes an
overview of the development of Kiryu with indicators of changes in production, labor
force, structure of firms, and technology. Section 3 examines the characteristic of
firms with the employment of more than 10 workers from 1895 to 1918, whose
production record was collected by various statistical surveys. Three hypotheses
regarding the dominant firms are presented through the comparison among
export-oriented large firms, WMCs, and newly emerged joint stock firms. Section 4
presents the methodology of regression analysis and examines the results. The last
section concludes by summarizing the main findings of the study and drawing
implications for future research.
2. An Overview of the Development of the Kiryu Weaving District
This section examines the production growth in Kiryu since the late 19th
century and investigates the changes in the extent of the inter-firm division of labor and
the adoption of power looms which are considered as the keys to the growth of Kiryu.
Through the observation of structural changes, we will identify the two distinct phases
of growth in Kiryu in the early 20th century.
5
2-1 Production growth in the early 20th century in Kiryu
Figure 1 illustrates the changes in real value of production, employment, and
labor productivity in Kiryu, using index (1895 = 100). In the early 1900s, the real
value of production shows upward trend: It was 3.5 million yen in 1904 but increased to
roughly 10 million yen in 1907. Since then, it had been stagnant at around 10 millions
yen until 1914. In contrast, it experienced rapid growth around the boom period of the
First World War from 1914 to 1919, which was 17 times increase during the mere
5-year period. Even though it is well known that the 1920s was the era of repeated
recessions or depressions in Japanese economy, surprisingly real value of production in
Kiryu was maintained subsequently at the level between around 60 to 70 millions yen
until 1929.
In Figure 1, solid and broken curves show the indices of the total number of
workers and female workers, respectively. Since the female workers account for 70 to
80 percent of labor force, the two curves look alike. From the late 1890s to 1900s, the
total number of workers decreased and dropped to around 7,000 in 1904. It continues
to stagnate through the late 1900s but begins to increase toward the end of the 1910s.
It was around 9,000 in 1910 and rose to 13,500 in 1920. It suddenly dropped to less
than half in 1921 due to depression. In the late 1920s, it finally began the recovery
6
process. It can be confirmed that the total labor force increased faster than female
labor force in the 1910s, implying that male employment grew faster than female
employment. It is also clear that the increase in production was not caused primarily
by the increased input of workers.
2-2 Smithian growth and Schumpeterian development
If we turn to the changes in labor productivity, it is apparent that it was
improvement of labor productivity that contributed to boosting the real value of
production. More importantly, it can be recognized that there are three distinct phases
of increase in labor productivity; (1) gradual growth in the 1900s, (2) stagnation from
the end of the 1900s to the mid-1910s, and (3) drastically rapid growth from the
mid-1910s to the mid-1920s. Average annual growth rate in labor productivity was
0.92% from 1904 to 1915 and 9.95% from 1916 to 1927.
Figure 2 examines the changes in the number of out-weavers and other
production organizations including factories, cottages, and WMCs in Kiryu.2 The
number of out-weavers, who are primarily based in surrounding villages around Kiryu
2 The data in 1904 and 1905 are not available. Statistical Survey of Gunma Prefecture where Kiryu was located defines four types of production organizations with two criteria, i.e. the numbers of workers and ownership of raw materials. Factory is defined as a workshop with more than ten workers and cottage as one with less than ten workers. On the other hand, the defining characteristics of the weaving manufacturer-cum-contractor is to put out raw materials to out-weavers. Out-weavers are those who are engaged in weaving for contractors. For details, see Hashino (2007a, p. 34, footnote 2).
7
town, increased rapidly from the mid-1900s to the 1910s. It was approximately 3,700
in 1906 but rose to 5,800 in 1914. Note that the average number of workers per
out-weaver workshop had been stable at about 1.5. In contrast, the number of other
production organizations continued to stagnate at around 500 until it shows increasing
trend from the mid-1910s. These observations clearly indicate the increase in the
number of out-weavers per other production organization. Indeed, it almost doubled
from 6.8 in 1906 to 11.6 in 1914. This indicates the expansion of division of labor,
which was organized by WMCs. This division of labor happened not only in weaving
process but also in many other processes carried out by specialized subcontractors
(Hashino and Kurosawa 2011).3
Figure 3 illustrates the coordination activities of WMCs (left) and specialized
processes carried out by subcontractors (right) in Kiryu around 1910. It is apparent
that many processes were carried out by specialized subcontractors. It must be noted,
however, that WMCs were originally engaged in the whole production processes but
gradually out-sourced many processes, such as throwing, dying, designing, weaving,
and finishing. For example, dyeing process was one of the key preparatory processes
3 Specialized subcontractors are generally small and located in Kiryu town. Whether their activities were recorded in statistics is not known in earlier period. Responding to a significant increase in the number of dyers, the prefectural government included the dyeing industry in its annual statistical survey after 1915 (Hashino and Kurosawa 2011).
8
carried by WMCs in the earlier period. Since the introduction of synthetic dyes in the
1880s, whose proper use required scientific knowledge, WMCs gave up dyeing and
began to put out the process to specialized subcontractors, who have acquired such
knowledge. Henceforth, division and specialization of labor were gradually and
widely diffused. Such an evolutionary process can be termed as Smithian growth.4
It is extremely important to note that despite Smithian growth from the late
1900s to the mid-1910s indicated by the increasing number of out-weavers (Figure 2),
neither total production nor employment increased significantly in this period (Figure 1).
As will be discussed in the next section, we attribute this puzzling observation to the
failure of large export-oriented firms and offsetting rise of WMCs in the early 20th
century. The former aggressively mechanized since the late 19th century to create
added value for exported products through improved designs, textures, and luster, but
failed to expand the production (Hashino and Kurosawa 2011).
Undoubtedly the most important single innovation in Kiryu was the
introduction of power looms, whose dissemination can be traced by the changing
proportion of power looms in this period. As is shown in Figure 4, the adoption of
4 Interesting statistical data collected by Kiryu Trade Association for Weaving indicates how widely outsourcing diffused in Kiryu. According to their report in 1900, there were 853 weaving producers, 37 fabric merchants, 16 scouring and finishing producers, 62 raw silk merchants, 18 dyers, 14 dyestuff merchants, 6 cotton-yarn merchants, 12 designers for jacquard machines, 25 reed producers, 115 warping producers, and 6,725 out-weavers within the district (Hashino and Kurosawa 2011).
9
power looms in the Kiryu district as a whole (bold curve) started to increase from the
mid-1910s and grew rapidly toward the 1920s. The average adoption rate was only
4.9% in 1915 but drastically increased to 84.1% in 1930.
Solid and broken curves in the same figure show the proportion of power
looms in Kiryu city (former Kiryu town) and Yamada county (surrounding rural
villages) within the Kiryu weaving district, respectively. It is interesting to observe
that the proportion of power looms in Kiryu city was much higher than that of Yamada
county already in 1921. Therefore, we can assume that the introduction of power
looms in urban area proceeded at much faster rate than that in rural villages, probably
even in the 1910s. New technology needed new production organizations, because the
use of power looms confers scale advantages. Previous studies report that the
introduction of power loom was accompanied by the adoption of factory system in
Japan (Hashino 2007a; Hunter 2003; Minami and Makino 1983; Saito and Abe 1988).
It is therefore likely that factories with power looms located in the center of the district
played an important role in promoting Schumpeterian development since the late 1910s
in Kiryu. No less important might be the establishment of joint stock firms which
contributed to financing large investments in factory buildings and machineries.5
5 Although we do not analyze in this study, firms which adopted power looms made a number of improvements in the product designs in order to produce traditional products by power looms.
10
To sum up, it seems legitimate to call the first trend of output growth Smithian
growth and the second one Schumpeterian development. Smithian growth is chiefly
caused by the increase in the division of labor which must have been created by
reduction in the transaction cost associated with the improved assignment of tasks and
enforcement of property right and production responsibilities (Mokyr 1990). On the
other hand, Schumpeterian development is derived from the major increase in
production efficiency by innovations. Such innovations encompass new production
technology and organizational changes that enable innovative firms either production of
a given output with much smaller amount of resources or the production of much better
or large quantity of new products with the same resources, or both (Mokyr 1990).
3. Changing Characteristics of Sample Firms and Hypotheses
3-1 Characteristics of sample firms
Based on the above discussions, we attempt to investigate the behaviors of
weaving firms in Kiryu with the employment of more than 10 workers covered by
Statistical Survey of Gunma prefectural government and Factory Survey.6 These firms
6 The data source in 1915 and 1918 is Factory Survey, which was conducted by Gunma prefectural government in order to report to the central government. Compared with Statistical Survey data used for 1906 and 1910, individual data in 1915 and 1918 contain much more detailed information about production of each firm.
11
were termed factories. Although there are four production organizations (i.e.,
out-weavers, factories with more than 10 workers, cottages with less than 10 workers,
and MWCs), it is possible that the increase in the number of workers converted the
cottages and MWCs to factories.
Table 1 exhibits the average characteristics of weaving firms in selected years
from 1895 to 1918. A glance establishes that the number of firms was only ten in the
late 19th century but tripled in the 1900s. Furthermore, the number more than doubled
in the early 1910s and reached 88 in 1918. Why did the number of firms with the
employment of more than 10 workers increase appreciably? The average year of
establishment of firms indicates that the entry of new large firms was not necessarily the
major reason. Recall that aside from out-weavers, there are three types of sample
weaving firms in Kiryu; (1) large firms which attempted ‘vertical integration’,7 (2)
traditional firms including WMCs whose number of workers increased to more than 10,
and (3) newly-established joint stock firms which equipped power looms and adopted
factory system in the late 1910s.
Large firms with the employment of nearly 100 workers seem to have appeared
7 In our study, vertical integration refers to the production system in which preparatory, weaving and finishing processes are carried out within a firm. Some of them were typically out-sourced processes in the case of WMCs as is shown in Figure 3. It is partly similar to that discussed by Jones (1987), i.e., the backward integration system in British silk industry in the 1820s and 1830s, in which throwing and weaving process were carried out internally.
12
in the late 19th century. The average year of establishment was 1851 in 1895 and 1876
in 1899, respectively, which are much different. In 1895, it seems that old firms were
dominant, even though there were a few newly-established firms which just started their
operation. Judging from the rising average year of establishment in 1899, it can be
considered that newly-established firms became dominant. The latter firms attempted
large-scale vertical integration with extraordinarily large western machines mainly for
preparatory and finishing processes. These firms did not depend on the division of
labor with other small firms, unlike WMCs, which means that they did not enjoy
agglomeration economies arising from inter-firm transactions. They used hand looms
except for Nihon Orimono Corporation, which tried to produce exportable products but
faced difficulties in operating large-scale factory (Kameda 2011).8
Interestingly, from 1899 to 1903 not only the average number of workers
sharply declined from 91.5 to 33.1 but also the average year of establishment changed
from 1876 to 1867. On the other hand, the number of firms became tripled between
the two years. Thus, it is clear that major players promoting the growth drastically
changed in this period. In other words, while the large-scale firms failed their business,
relatively old WMCs became dominant in the 1900s. The average number of workers
8 According to Kameda (2011), this company installed imported power looms.
13
continued to decline to 23.2 in 1906 (see column of 1906b in Table 1, which excludes
Nihon Orimono Cooperaton, as it is outlier). Female worker ratios in 1903 and 1906
were also lower than those in the 1890s, which strongly indicates that the relatively
large number of male workers who were engaged in preparatory processes and delivery
of yarns to out-weavers increased. Loom/worker ratio was far less than unity in 1906,
which means that not all workers in the weaving firms were engaged in the weaving
process. From the above discussions, it seems clear that some WMCs grew to be
medium-scale firms and promoted the growth of Kiryu by expanding division of labor.
It is likely that the stagnant production from the mid-1900s to the mid-1910s observed
in Figure 1 was the result of the reduction in production by large firms and offsetting
increase in production by growing WMCs.
In the 1910s, new entries can be recognized from both increase in the number
of firms and the rise of the average year of establishment. While the average number
of workers slightly increased compared with the 1900s, female worker ratio continued
to decline. WMCs would have been still dominant in this period but newly entering
firms gradually expanded their scale of operation. From 1915 to 1918, the average
number of firms and the number of workers rapidly increased with concomitant rise in
the average year of establishment, which indicate increase in the number of
14
newly-established large firms. At the same time, the proportion of power loom
reached 80%. Such newly established large firms which appeared in the 1910s can be
considered as the new major players which promoted Schumpeterian development. It
is worth pointing out that with the advent of such firms, the average sale revenue per
firm increased six times from 1915 to 1918. Note that since the cost of putting-out
contracts and out-sourcing are not included in the sales revenue, it becomes larger as the
extent of the division of labor is larger.
3-2 Comparison of Export-Oriented Firms with Other Firms
What kind of products did our sample firms produce? Some of their
products were shipped for domestic markets but others were for export markets.
Organized by WMCs, traditional products for domestic market were produced by
utilizing complicated and specialized division of labor within the district, as was
illustrated by Figure 3. In contrast, the large firms established in the late 19th century
attempted to sell product at export markets without depending on any division of labor.
In order to compare the firms with different market orientations and locations,
Table 2 undertakes the comparison of export-oriented firms with other domestic
market-oriented firms in Kiryu town and outside in 1906, 1910, and 1915.9 Data
9 We estimate the export- and domestic market-orientation from the main product reported by the survey.
15
source and sample size are the same as in Table 1. The reason why we regard location
as important is that leading WMCs tended to be located in Kiryu town partly because
this is most convenient for them to organize putting-out contracts in various processes
and partly because access to railways and electricity was also favorable in town.
According to Table 2, however, other firms outside Kiryu town were likely to be WMCs,
because, although smaller, their sales revenue was not low in 1915, if we consider the
smaller number of employed workers. The number of export-oriented firms increased
from 1906 to 1910 and then declined, whereas the number of other firms in both Kiryu
town and outside continued to increase and became dominant in 1915. This suggests
that the second players, large WMCs, supported Smithian growth.
In terms of the number of workers, export-oriented firms were largest in all
years but their employment size declined from 71.8 workers in 1906 to 45.4 workers in
1910. Judging from the facts that the number of export-oriented firms doubled from
1906 to 1910 and that average year of establishment in 1910 is 6 years younger than
that in 1906, large firms disappeared and the newly established firms with more
moderate scale became dominant.
Female worker ratio tends to decline over time but it is much lower in other
firms in Kiryu town. The relatively large number of male workers was employed by
16
other firms in Kiryu town because they are likely to be leading WMCs type firms, in
which the relatively larger number of male workers played an important role in the
preparatory processes, delivery of raw materials, and collection of finished products.
Percent of firms using traditional water wheels was also high among other firms in
Kiryu town because they were used for preparatory processes (Hashino 2007c). Only
some export-oriented firms introduced motive-powers in 1906 and 1910, which was
steam engine. In 1915, however, some of other firms in Kiryu town also equipped
powers, even though their adoption rate was lower than that of export-oriented firms.
This is likely because electric powers were supplied by the Watarase Water Power
Electricity Company, which was established in 1906 and started operation in 1908
(Kiryu Orimonoshi Hensankai 1940). Prior to supply of electricity, export-oriented
firms had to equip motive-powers such as steam engines on their own account. It must
be also pointed out that the number of workers of the export-oriented firms far exceeded
the number of looms in 1906, which indicates that many workers were engaged in a
variety of production activities other than weaving within the large factories.
Until the late 1900s, hand looms were used in almost all firms in Kiryu except for
a few (Kiryu Orimonoshi Hensankai 1940). It is therefore doubtful whether the
large-scale export-oriented firms could enjoy scale advantage in the absence of large
17
fixed inputs. In fact, such large-scale firms disappeared and were replaced by smaller
firms with the employment of much less than 50 workers, which were likely to be
successful WMC-type firms whose advantage lay in the use of the division of labor with
out-weavers and other supporting firms. Therefore, we advance the following
hypothesis:
Hypothesis 1: Although more than several large firms with the employment of
more than 50 workers and the use of hand looms were founded in the late 19th
century, they soon collapsed due, at least partly, to the lack of scale advantages. In
contrast, WMCs thrived based on the out-weaving system in the beginning of 20th
century in Kiryu.
3-3 Comparison of Newly Emerging Large Firms with Other Firms
Our sample firms in 1918 benefitted from the economic boom during the First
World War (1915-1918). Responding to the increasing demand for the products in
Kiryu, the number of large firms increased. Wage rates also increased sharply in Japan,
surpassing the Lewisian turning point according to Fei and Ranis (1963). In fact, in
local labor markets in the neighborhood of Kiryu, wage rates of female workers in
weaving, silk-reeling, and farming sharply increased in this period (Hashino 2007a).
Hence many weaving firms in Kiryu used power looms to save labor by using electricity.
18
Establishment of new factories and workshops as well as the installation of power
looms would have required large investment funds, which seems to have led to the
establishment of joint stock firms, as will be shown shortly.
Table 3 analyzes the characteristics of 88 firms in 1918 from the perspective of
market orientation and location. Joint stock firms were export-oriented and
established around the war boom period. They were particularly large with the
employment of more than 300 workers, which clearly indicates that they sought the
scale advantages. There are many differences between joint stock firms and other
firms within the category of export-oriented firms. Caution is needed in interpreting
number of looms because there were both hand and power looms. In the case of joint
stock firms, female workers operated primarily power looms, whereas in other
export-oriented firms female workers used both hand and power looms. It must be
emphasized that most joint stock firms were subcontractors for smaller weaving firms
and received fees of undertaking preparatory and finishing activities from other firms
within the districts. Such behavior is consistent with the theory of the division of labor
formulated by Stigler (1951), which argues that one of the main sources of the division
of labor is the different optimum scales of production in different sub-production
processes. In this respect they are far different from the large export-oriented firms
19
established a few decades earlier. Based on the above discussion, therefore, we
postulate the following hypothesis regarding the large firms which emerged in the
mid-1910s.
Hypothesis 2: Several large joint stock firms founded in the mid-1910s were
Schumpeterian innovators, who transformed Kiryu weaving district by realizing
the scale advantages associated with the introduction of power looms, factory
production, and electricity.
Let us turn to the characteristics of domestic market-oriented firms.
Remarkable differences between 32 firms in Kiryu town and 15 others can be observed
in female worker ratio, the number of looms, use of electricity, and holding of registered
trademarks in the 1900s. The formers were likely to be large WCMs employing
relatively many male workers without much internal production. In the case of a
leading WCM, called Goto Firm, about which Hashino (2007a) explores the relation
between adoption of new production organization and technology choice, increase in
male workers in the mid-1910s was associated with the expansion of out-weavers. Its
strategy was low volume production of a wide variety of products for domestic market
by enjoying the advantage of agglomeration economies.10 In addition, holding of
10 For the flexibility of production, Nakabayashi (2007) also admits that putting-out system using hand looms had advantage in Kiryu until the 1910s.
20
registered trademarks in Meiji Era indicates that other firms in Kiryu town were old
leading firms11. They also began the introduction of power looms, but percentage of
power looms were the lowest among the four groups in Table 3. For example, the
introduction of power looms in the above mentioned Goto Firm was delayed and
occurred from the late 1910s to 1920s. This accompanied adoption of factory
production system and giving up of producing a wide variety of products unsuited for
mechanized mass production. Therefore, we advance the following hypothesis
regarding WCMs:
Hypothesis 3: WCMs became laggards in the era of the Schumpeterian innovation.
4. Econometric Analyses
In order to test the validity of the hypotheses postulated in previous section, in
this section we estimate the regression functions explaining the number of workers,
female worker ratio, the number of looms, loom/worker ratio, proportion of power loom,
sales revenue, and sales/worker ratio in 1906, 1910, 1915, and 1918. Note that not all
the data of dependent variables are available in every year except in 1918. Sample
firms are also different from year to year, so that the panel data analysis cannot be
11 According to Arai (year unknown), around 100 trademarks were registered by firms, which are considered mainly as WMCs, from the 1890s to the 1900s.
21
applied. Thus, we estimated the regression functions separately in each year.
4-1. Specification of Regression Functions
Denoting the dependent variables mentioned above by Y, the estimated regression
function is specified as follows:
Yit = α0t + α1t(Edo period dummy)i + α2t(Operation years in Meiji era)it +
firm dummy)it + α6t(Power use dummy)it+ α7t(Wheel use dummy)it+
α8t(Trademark dummy)i+ εit ,
where subscripts i and t refer to i-th firm and t-th year, respectively; αs are regression
parameters; and ε is an error term. Edo period establishment dummy and the number
of operation years in the Meiji era for those firms established in the Meiji period are
used to examine if the experience of weaving business affects the scale of operation and
productivity. We use three mutually exclusive firm dummies, in which the base of
comparison is domestic market-oriented firms in Kiryu town (most likely WMCs):
“Export firm dummy” refers to export-oriented firms in 1906, 1910, and 1915 and to
export-oriented firms other than joint stock firms in 1918; “Domestic firm dummy”
means domestic market-oriented firms located outside Kiryu town;12 and “Join stock
12 Since there was only one such firm in 1906, it was combined with domestic market-oriented firms in Kiryu town in the regression analysis.
22
firm dummy” was used only for 1918 regression because they were too few in
previous years. The dummy for trademarks registered in 1897-1907 period, which was
used only in 1918 regression, is expected to capture the behavior of the leading
WMCs.13
Problematic as explanatory variables are export firm, domestic firm, and joint
stock firm dummies, and power use and water wheel use dummies, as they are likely to
be endogenous. Due to the paucity of exogenous variables, however, we are forced to
use them as explanatory variables. To the extent that they are positively correlated
with unobservable factors included in the error term, such as managerial abilities, their
estimated coefficients tend to be over-estimated. Thus, we can hardly assert the
causality from the estimated coefficients of these variables. What can be conjectured
is association or the correlation of the variables of our interest. Another caveat is that
while we apply the ordinary least squares regression when depended variables are
continuous, we apply the tobit estimation method when dependent variables are
truncated, such as the number of looms and loom/worker ratio, which include zeros.
Since export orientation is expected to be positively associated with the scale
of operation, particularly in early years, the coefficient of export firm dummy (α3) is
13 We use ‘trademarks registered by firms in the Meiji 30s (from the late 1890s to the mid-1900s)’ shown in Arai (year unknown).
23
expected to be positive and significant in 1906 but becomes insignificant or less
significant in the regression equations dealing with the scale of operation in later years,
if Hypothesis 1 is correct. On the other hand, we expect the coefficient of joint stock
firm dummy (α5) to be positive and significant in the employment, the number of looms,
and revenue functions in 1918, if Hypothesis 2 is correct. Finally we expect the
coefficients of all three firm dummies in the proportion of power loom function to be
positve, as WMCs were inactive in the introduction of power looms according to
Hypothesis 3.
4-2. Estimation Results
Table 4 shows the estimation results for 1906, 1910, and 1915. Several
important observations can be made. First, neither coefficients of Edo period dummy
nor those of the operation years in the Meiji era are significant in any regression
equations.14 These coefficients are not significant, either, for 1918 to be shown in
Table 5. These findings indicate that the mere production experience did not affect the
performance of weaving firms. Second, export firm dummy is significant in all the
four regression functions in 1906, it becomes insignificant in the regression of the
number of workers in 1910 and 1915, and its coefficient is negative and significant in
14 The results remain qualitatively unchanged, even if we excluded the operation years in the Meiji era.
24
the sales per worker regression in 1915. The last result strongly suggests that sales
revenue per worker was significantly larger for WMCs, because they use out-weavers
and other sub-contractor. On the other hand, its impact on female worker ratio is
significantly positive in all three years, suggesting that female workers were employed
to operate large lots for producing export products. Note that the magnitude of the
coefficients of export dummy in the regression of the number of workers are not so
different among 1906, 1910, and 1915, even though those in 1910 and 1915 are
insignificant, which indicates the larger variations of employment size in the
export-oriented firms in these latter two years. Thus, it seems clear that the
export-oriented firms chose large-scale factory production system initially, while
employing relatively large number of female workers, but they failed to realize and
maintain scale advantages, as is reflected in its insignificant effect on labor employment
and sales revenue and even negative and significant effect on sales revenue per workers
in 1915. Such results are substantially different from the case of joint stock companies
to be examined from Table 5. The other side of the same coin is that leading WMCs
prospered in this period. These results are consistent with Hypothesis 1.
Third, power use dummy has significantly positive effects on the number of
workers in all three years, the number of looms in 1906, and sales revenue in 1915.
25
Note that the source of the power was steam in 1906 and 1910, but it was almost
completely replaced by electricity in 1915 (see Table 2). Also note that since only
large export-oriented firms used the steam power in 1906 and 1910, the combined
effects of export-orientation and the use of steam power were extremely large. Thus, it
appears that large export-oriented firms attempted to enjoy scale economies by adopting
the vertically integrated production system with the installation of large steam-power
generators. Since the magnitudes of coefficients of power use dummy in the
regression of the numbers of workers are similar among the three years, it is doubtful
that the replacement of steam by electricity brought about large changes in the
employment practice immediately. Fourth, dummy for domestic market-oriented firms
outside Kiryu town is insignificant, which indicates that the behaviors of WMCs in
Kiryu town and outside were not substantially different in 1915. Finally, it must be
pointed out that three of the coefficients of water wheel use dummy are positive and
significant in 1906, suggesting that in the absence of electricity, water wheel was used
to expand the scale of operation and adopt the capital-intensive production method not
for weaving but for other production processes.
Table 5 exhibits the estimation results of regression functions in 1918.
Interestingly, joint stock firm dummy is positively correlated with the number of
26
workers, number of looms, and sale revenue, as well as female worker ratio and
worker-loom ratio. Moreover, the coefficients in the regression functions for the
number of workers, number of looms, and sales revenue are comparatively large, which
strongly indicates that newly established large joint stock firms sought the scale
economies. In fact, if we compare joint stock firms and other export-oriented firms
shown in Table 3, revenue of the former, on average, exceeded the latte by 16 times,
whereas the number of workers is 12.5 times and the number of hand-loom equivalent
looms 13.4 times,15 indicating the strong scale advantages of the former over the
latter.16 These results support Hypothesis 2.
Somewhat unexpectedly, the effect of joint stock firm on the power loom ratio
is insignificant. This is because all the joint stock firms used powers and, hence, their
use of power looms is captured by the power use dummy. Although the difference is not
statistically significant, it is noteworthy that coefficient of joint stock firm dummy is
smaller that that of non-joint stock firm dummy in the loom/worker ratio regression.
This is likely because joint stock firms served as sub-contractors for other firms by
carrying out preparatory and finishing processes, which did not use looms. This may
15 According to Hashino (2007c), price of power loom was around 300 yen, whereas that of hand loom was 2 yen to 5 yen. Thus, the price ratio was 60 to 150. 16 Note that these firms did not contract out sub-processes, so that the revenue was generated by own production activities.
27
result in the negative and larger coefficient of joint stock firm dummy in the sales per
worker regression. This negative coefficient as well as the negative and significant
coefficient of non-joint stock export-oriented firm dummy in the sales per worker
regression indicates that WMCs in Kiryu town achieved significantly large sales
revenue per own worker due to the outsourcing to out-weavers and specialized
subcontractors.
The effects of non-joint stock firm dummy are similar to those of joint stock
firms except for the size effects on the number of workers and looms, and sales revenue.
Moreover, the magnitudes of its coefficients in the female worker ratio, loom/worker
ratio, and sales per worker regressions in Table 5 are similar to those in Table 4. It
appears that non-joint stock firms are not significantly different from the
export-oriented firms analyzed in Table 4.
The results examined so far imply that WMCs employed smaller number of
workers, used fewer looms, particularly power looms (Table 3), and lower sales revenue
than joint stock firms but attained much higher revenue per worker due to the
subcontracts with out-weavers and other specialized sub-contractors. WMCs in Kiryu
town, however, were different from domestic market-oriented firms outside Kiryu town:
WMCs in Kiryu town employed relatively more labor and less power loom (Table 5).
28
This is consistent with Hypothesis 3 that WMCs were less active in the introduction of
power looms, mostly likely because of their established and at least formerly successful
subcontract-based production systems
As may be expected, power use dummy is particularly significant in the power
loom ratio function. As Minami (1977) emphasizes, the electrification promoted the use
of motor-driven machines in Japan, which led to rapid decentralized industrialization in
the early 20th century Japan.
Although this study does not analyze the development of this weaving district
in the subsequent periods, it is known that WMCs actually followed the factory
production system introduced by the joint stock firms and used power looms, which
meant the demise of the out-weaving systems in favor of factory systems in this
weaving district (Hashino 2007a).
5. Conclusion
In this study, we focused on the performances of three types of firms
which contributed to the weaving production in Kiryu in the early 20th
century. Changing phases of production were characterized by the concepts
of Smithian growth and Schumpeterian development. Expansion and
29
sophistication of division of labor supported Smithian growth, in which
domestic market-oriented WMCs played a major role. Introduction of
power looms and factory system were the major drivers of Schumpeterian
development, which was promoted by large, export-oriented joint-stock firms.
Three hypotheses regarding the performance of the three types of firms were
tested by regression analyses, which reveal strategies of WMCs to utilize
sub-contracts, the pursuit of scale economies by export-oriented firms, and
the importance of electricity to facilitate successful adoption of power-looms
and factory systems. Although the rise of wage rates in the 1910s would
have affected the introduction of power looms and relative advantage of
factory systems, our study cannot identify its effect due to the cross-section
nature of our data sets.
We would like to conclude this study by identifying two major
remaining issues for further research. The first one is concerned with the
importance of Smithian growth. Of course, division of labor in various
industrial clusters were widely observed in developing economies,
particularly in the early stage of cluster development (Sonobe and Otsuka
2006, 2011), but its role has not been highlighted. In the case of Kiryu,
30
WMCs are worthy of being called entrepreneurs promoting Smithian growth
by organizing specialized production systems. Historically, they were
engaged not only in manufacturing but also in marketing, designing, quality
control, and making trial samples (Uchida 2002). In other words, WMCs
played the role of traders, who are essential for linking producers with
markets. Although they were not so keen about breakthroughs, they
effectively utilized existing resources in local community, e.g., cheap rural
labor, based on mutual trust (Hayami and Godo 2005). When the trust was
insufficient in mobilizing collective action, support of local trade association
or local government became crucial for strengthening agglomeration
economies (Hashino and Kurosawa 2011). It seems worth exploring the
extent to which Smithian growth lays foundation for the development of
industrial districts in its early stage of the development
Another issue is to explore how Schumpeterian development emerges
or what types of entrepreneurs play a role of Schumpeterian innovators.
Three types of innovations were observed in our study site; process
innovation (e.g., introduction of power looms), product innovation (e.g.,
introduction of new products for export), and organizational innovation (e.g.,
31
introduction of factory systems). Establishment of joint stock firms and
utilization of division of labor also fall under the organizational innovation.
Since the introduction of power looms and suitable products for mechanized
production, the adoption of factory systems, and the emergence of joint stock
firms are so closely interrelated with each other that they took place
simultaneously. Thus, successful innovations seem to require managerial
capacity to carry out a variety of component innovations. What type of
human capital is needed for such innovations must be clarified through
further historical research and research on the contemporary development of
industrial clusters in developing countries.
32
Acknowledgement
Earlier version of this paper was presented at Zushi Conference at University of Tokyo,
29th January, 2011. We would like to thank Professors Tetsuji Okazaki, Yasuyuki
Sawada, Minoru Sawai, and Tetsushi Sonobe for their valuable comments for
stimulating our study. First author wishes to thank Professor Janet Hunter and other
colleagues in Department of Economic History, London School of Economics and
Political Science, for welcoming her as a visiting fellow in 2011. This research was
supported by Rokkodai Koenkai at Kobe University, Grant-in-Aid for Scientific
Research (A) 23243022, and (A) 23243055.
References
Abe, T. (1992) “The Development of the Producing-Center Cotton Textile Industry in
Japan between the Two World Wars,” Japanese Yearbook on Business History, 9,
4-27.
Abe, T. (1999) “The Development of the Putting-Out System in Modern Japan: The
Case of the Cotton-Weaving Industry,” in Odaka, K. and Sawai, M. (eds.) Small
Firms, Large Concerns (New York: Oxford University Press), 217-219.
Abe, T and Saito, O. (1988) “From Putting-out to the Factory: a Cotton Weaving
33
District in Late-Meiji Japan,” Textile History 19(2), 117-142
Arai, K. (year unknown) Kiryu Orimono Toukei Shoshiryo 2 [Statistics and Materials
of Kiryu Weaving 2] (not for sale).
Fei, J.C.H., and Ranis, G. (1963) “Innovation, Capital Accumulation, and Economic
Development,” American Economic Review 53 (3), 283-313.
Fujita, T. (1998) “Local Trade Associations (Dogyo Kumiai) in Prewar Japan,” in
Yamazaki, H. and Miyamoto, M. (eds.) Trade Associations in Business History
(Tokyo: University of Tokyo Press), 87-113.
Hashino, T. (2007a) “The Rise of Power-Loom Factory in the Traditional Silk-Weaving
District,” in Okazaki, T. (ed.) Production Organization in Japanese Economic
Development (London: Routledge), 14-38.
Hashino, T. (2007b) “Industrial District, the Rise and Growth of Small and
Medium-Sized Enterprises: Mechanical Engineering in the Higashi-Mikawa
District,” Japanese Research in Business History 24, 53-75.
Hashino, T. (2007c) Keizai Hatten to Sanchi, Shijo, Seido: Meijiki Kinu Orimonogyo no
Shinka to dainamizumu [Economic Development, Industrial Districts, Markets,
and Institutions: the Evolution and Dynamics in the Silk-Weaving Industry in
Meiji Period] (Kyoto: Minerva).
34
Hashino, T. (2010) “The Importance of Consular Reports for the Export Growth of
Japanese Silk Fabric Habutae in the Late 19th Century,” in Ulbert, J. and Prijac,
L. (eds.) Consulship in the 19th Century (Hamburg: Dobu), 485-501.
Hashino, T. (forthcoming) “Institutionalising Technical Education: The Case of
Weaving Districts in Meiji Japan,” Australian Economic History Review.
Hashino, T and Kurosawa, T. (2011) “Beyond Marshallian Agglomeration Economies:
The Role of the Local Trade Association in a Meiji Japan Weaving District
(1868-1912),” Discussion Paper Series No.1113, Graduate School of Economics,
Kobe University.
Hayami, Y. and Godo, Y. (2005) Developing Economics: From the Poverty to the Wealth
of Nations (New York: Oxford University Press).
Hunter, J. (2003) Women and the Labor Market in Japan’s Industrialising Economy:
The Textile Industry before The Pacific War (London: Routledge).
Jones, S.R.H. (1987) “Technology, Transaction Costs, and the Transition to Factory
Production in the British Silk Industry, 1700-1870,” Journal of Economic
History 47(1), 71-96.
Kameda, M. (2011) “Nihon Orimono Gaisha no Gijutsu to Keiei” [Technology and
Management of Japan Weaving Company], Kiryu Shien 50, 1-29.
35
Kiryu Orimonoshi Hensankai [Editorial Association of the History of Kiryu Silk
Weaving Industry] (1940) Kiryu Orimonoshi 3 [History of the Kiryu Weaving
Industry 3] (Kiryu: Kiryu Trade Association for Weaving).
Marshall, A. (1920) Principles of Economics (London: Macmillan, now Palgrave
Macmillan).
Minami, R. (1977) “Mechanical Power in the Industrialization of Japan,” Journal of
Economic History 37 (4), 935-958.
Minami, R. and Makino, F. (1983) “Conditions for technological diffusion: case of
power looms,” Hitotsubashi Journal of Economics 23 (2), 1-20.
Mokyr, N. (1990) The Lever of Riches: Technological Creativity and Economic
Progress (Oxford: Oxford University Press).
Nadvi, K. (1999) “The Cutting Edge: Collective Efficiency and International
Competitiveness in Pakistan,” Oxford Development Studies 27(1), 81-107.
Nakabayashi, M. (2007) “Flexibility and Diversity: The Putting-out System in the Silk
Fabric Industry of Kiryu, Japan,” in Okazaki,T (ed.), 128-166.
Parker, W. N. (1984) Europe, America, and the Wider World: Essays on the Economic
History of Western Capitalism (Cambridge: Cambridge University Press).
Piore, M. J., and Sabel, C. F. (1984) The Second Industrial Divide: Possibilities for
36
Prosperity (New York: Basic Books).
Porter, M. E. (1998) On Competition (Boston: Harvard Business School).
Sabel, C. F., and Zeitlin, J. (1999) World of Possibilities: Flexibility and Mass
Production in Western Industrialization (Cambridge: Cambridge University
Press).
Sawai, M. (1999) “The Role of Technical Education and Public Research Institutes in
the Development of Small and Medium Enterprises: The Case of Osaka between
the Wars,” in Odaka, K. and Sawai, M. (eds.), 250-89.
Schmitz, H. (1995) “Collective Efficiency: Growth path for Small-Scale Industry,”
Journal of Development Studies 31 (4), 529-566.
Sonobe, T. and Otsuka, K. (2006) Cluster-Based Industrial Development: An East Asian
Model (New York: Palgrave MacMillan).
Sonobe, T. and Otsuka, K. (2011) Cluster-Based Industrial Development: A
Comparative Study of Asia and Africa (New York: Palgrave MacMillan).
Stigler, G. J. (1951) “The Division of Labor is Limited by the Extent of the Market,”
Journal of Political Economy 59 (3), 183-193.
Tolliday, S. and Yonemitsu, Y. (2007) “Microfirms and Industrial Districts in Japan: The
Dynamics of the Arita Ceramic-Ware Industry in the Twentieth Century,”
37
Journal of Japanese Studies 33 (1), 29-66.
Uchida, H. (2002) “Shonin no Seisanteki Kino ni tsuite,” [Manufacturing Functions of
Traders], Bulletin of Museum of Commercial History, Osaka University of
Commerce 2, 1-16.
Zeitlin, J. (2009) “Industrial Districts and Regional Clusters,” in Zeitlin, J. and Jones, J.
(eds.) The Oxford Handbook of Business History (Oxford: Oxford University
Press), 219-243.
38
Figure 1. Changes in real value of production, employment, and labor productivity in Kiryu, 1895-1930 in semi-log scale (Index, 1895=100)
1.0
10.0
100.0
1000.0
10000.0
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
year
production total workers female workers labor productivity
Source: History of Weaving Industry in Gunma Prefecture [Gunmaken Orimono Enkaku Chosasho] (1895-1901); Statistical Survey of Gunma Prefecture
[Gunmaken Tokeisho] (after 1902).
Note: For realized value, we used price index for textile products in Ohkawa, K. et al. Estimates of Long-term Economic Statistics on Japan since 1868
